电子与信息学报
電子與信息學報
전자여신식학보
JOURNAL OF ELECTRONICS & INFORMATION TECHNOLOGY
2014年
6期
1485-1491
,共7页
视频推荐%行为分析%权重增量%Apriori算法
視頻推薦%行為分析%權重增量%Apriori算法
시빈추천%행위분석%권중증량%Apriori산법
Video recommendation%Behavior analysis%Incremental weight%Apriori algorithm
该文采用权重增量及相似聚集的用户行为分析算法,为用户推荐个性化视频提供了一个有效的解决方案。方法包含3个主要部分,首先利用RFM(Recentness, Frequency, Mon etary amount)模型分析用户的行为,将相同行为的用户归为一组;然后结合用户的最近习惯,使用基于权重增量的Apriori算法挖掘用户之间的关联规则,并用向量空间模型进行相似度计算从而实现用户相似聚集;最后进行协同过滤式推荐,完成整体个性化视频推荐过程。该方法的特点是行为数据自动收集获取,避免了直接对视频大数据的处理;另外,视频推荐随着用户行为的改变而动态变化,更加符合实际情况。实验结果表明,该方法有效并且稳定,相比于单一推荐方法,在准确率、召回率等综合指标上均有明显提升。
該文採用權重增量及相似聚集的用戶行為分析算法,為用戶推薦箇性化視頻提供瞭一箇有效的解決方案。方法包含3箇主要部分,首先利用RFM(Recentness, Frequency, Mon etary amount)模型分析用戶的行為,將相同行為的用戶歸為一組;然後結閤用戶的最近習慣,使用基于權重增量的Apriori算法挖掘用戶之間的關聯規則,併用嚮量空間模型進行相似度計算從而實現用戶相似聚集;最後進行協同過濾式推薦,完成整體箇性化視頻推薦過程。該方法的特點是行為數據自動收集穫取,避免瞭直接對視頻大數據的處理;另外,視頻推薦隨著用戶行為的改變而動態變化,更加符閤實際情況。實驗結果錶明,該方法有效併且穩定,相比于單一推薦方法,在準確率、召迴率等綜閤指標上均有明顯提升。
해문채용권중증량급상사취집적용호행위분석산법,위용호추천개성화시빈제공료일개유효적해결방안。방법포함3개주요부분,수선이용RFM(Recentness, Frequency, Mon etary amount)모형분석용호적행위,장상동행위적용호귀위일조;연후결합용호적최근습관,사용기우권중증량적Apriori산법알굴용호지간적관련규칙,병용향량공간모형진행상사도계산종이실현용호상사취집;최후진행협동과려식추천,완성정체개성화시빈추천과정。해방법적특점시행위수거자동수집획취,피면료직접대시빈대수거적처리;령외,시빈추천수착용호행위적개변이동태변화,경가부합실제정황。실험결과표명,해방법유효병차은정,상비우단일추천방법,재준학솔、소회솔등종합지표상균유명현제승。
This paper presents an effective solution for personalized video recommendation based on the weight increment and similar aggregation user behavior analysis algorithm. The method is implemented in three steps:first, the user behavior is analyzed using the RFM (Recentness, Frequency, Monetary amount) model, users with the same behavior are classified as a group;second, the Apriori algorithm based on weight increment is applied to mining association rules between users in line with the recent habits of users, and by using the VSM model for similarity calculation, the user similarity aggregation is realized; finally, the whole process of personalized video recommendation is completed by means of collaborative filtering. The proposed method can automatically collects user behavioral data and avoids direct video big data processing. In addition, the video recommend dynamically changes with the change of user behavior. The experiment results show that, the presented effective and stable, and the method achieves significantly increasement in precision and recall comparing with the single recommendation method.